Where’s best to give birth? Easy question, tough to answer

Note: the authors of the Birthplace Study have taken serious issue with this analysis, and I have posted their comments at the end. For ease of reference I have annotated in bold the claims they rebut and listed the rebuttals by number in the comment, which is by Dr Jennifer Hollowell, Prof Peter Brocklehurst, Elizabeth Schroeder and Prof Stavros Petrou on behalf of the Birthplace co-investigators.

The Birthplace Study, published in November last year, was intended to answer long-standing questions about the safest place for women to give birth in the UK – at home, in obstetric units in hospital, in freestanding midwifery units, or in midwifery units located on the same site as an obstetric unit (ie, in hospital).

There is little secret that the Government, which has offered women choice of the place of birth, would prefer them to choose cheaper options – midwifery units rather than hospitals – and the results have been presented by some as justifying this preference. (1)

A paper published in the current issue of BMJ shows what the cost difference amounts to (the full online version, published a couple of weeks ago, is here, or in a longer form, as Report 5 on the Birthplace Study website). A team led by Elizabeth Schroeder of the National Perinatal Epidemiology Unit at Oxford reports that births at home are cheapest - £565 cheaper than birth in an obstetric unit for all women at low risk of complications. The corresponding figure for freestanding maternity units is £196, and for maternity units in hospitals is £170.

These are quite small differences - smaller still when you look at women having their first babies. And while costs are less, adverse outcomes are more common in home births and births in maternity units of either sort for first-time mothers. (2) This leaves some uncertainty over which is truly the best option. For example, there is a 0.63 per cent probability of home birth and a 0.35 probability of a free-standing maternity unit being the most cost-effective option for low-risk women having their first babies if one sets the cost of an adverse event at £20,000.

The flaw with this study is its short timeframe. The data was limited to adverse outcomes at the time of birth, with no long-term follow-up. Accidents during birth are one of the costliest misadventures it is possible to imagine, so much so that private insurers will not insure against it. A child suffering from cerebral palsy can be disabled for life, with costs running into tens of millions of pounds. Yet the study looks only at immediate costs, which it acknowledges as a limitation.

I’d go further and suggest that it invalidates the whole comparison. A clue is provided in Figure 32 (below), which plots cost effectiveness curves for various thresholds for women having their first baby without complicating conditions at start of labour. At cost thresholds of £60,000 and over, obstetric units (hospitals) have the greatest probability of being the right place for such women to give birth.


So the “right” place is determined by the cost attributed to adverse events and this study cannot calculate what that is. Cost-effectiveness calculations are unhelpful if you leave out what is likely to be the biggest cost. The Birthplace Study couldn’t measure this, of course. But by publishing this paper it may give some people the impression that the answer is clear when it isn’t. (3)

The National Institute for health and Clinical Excellence (NICE) is drawing up new guidance on maternity care. Professor Mark Baker, head of clinical practice for NICE, told The Daily Telegraph on Saturday that the guidelines would be based on the best available evidence. The Birthplace Study, which cost £12 million, will certainly be included.

But it is not without its critics, of which the Birth Trauma Association has been the most persistent. The Royal College of Obstetricians and Gynaecologists also raised some issues, suggesting that first-time mothers should be advised of the benefits of obstetric units and midwifery units based in hospitals. For mothers having subsequent children, the place of birth did not appear to affect the outcome.

The BTA has a number of criticisms. They include loss of data, particularly from the freestanding maternity units. With adverse events being so infrequent, it was essential to capture all the data, but the team did not manage to do so. In the case of the freestanding maternity units, where only two thirds returned more than 85 per cent of the data properly, the “lost” data is likely to refer to women transferred from FMU’s to obstetric units because of complications. If so, the data loss flatters the FMUs and makes the obstetric units results appear worse. Failing to record this data cannot be blamed on overwork, since most FMUs have a relatively low throughput.  (4)

Nor is it clear that the team was comparing like with like. Those who gave birth in obstetric units had more complicating conditions at the start of labour, and more of them had other health conditions, too. Nearly 20 per cent had serious problems at the start of labour, far more than in the midwifery units. So the comparison drawn – “no difference in outcomes between midwifery units and obstetric units” – is not a fair one. (5)

If you exclude women with complicating conditions at the start of labour (Table 59, in Report 4 on the Birthplace Study website) the units which returned their data properly did show a  statistically significant difference. Adverse events are more than twice as common in FMUs (odds ratio 2.29, 95 per cent CI 1.17 to 4.47) as in obstetric units for women having their first baby. (6)

What of women transferred from freestanding maternity units to obstetric units, either before or after labour? If things go wrong, a transfer to hospital is needed, which may be lengthy if the maternity unit is not inside a hospital already. Such transfers are very common in women having their first baby: 45 per cent of planned home births were transferred during labour or immediately after birth, and 36 per cent of those in freestanding maternity units. The cost-effectiveness study shows that transfers from home took on average 29 minutes, from freestanding units 35 minutes, and from in-hospital maternity units 10 minutes.    

Table 31 shows strikingly poor results for women who made this transfer either from home or freestanding units after labour, with more than eight times as many suffering adverse events as those who transferred from maternity units within hospitals. The report does not comment on this in its conclusions. (7)

Among the points made by the RCOG is why, in this low-risk population of mothers, 20 of the 32 deaths were in the home or freestanding maternity unit group (Table 48). The BTA also finds it surprising that no deaths among mothers are recorded at all. Such deaths, though rare, occur at a rate of around one per 12,000, so it is indeed surprising – though not impossible - that in 80,000 births none were recorded. However, these were low-risk pregnancies. (8)

The rapid responses to both studies in the BMJ are worth reading. Generally, midwives are supportive, doctors sceptical, and campaigners for safe childbirth such as the BTA and Pauline Hull, a pro-Caesarean campaigner and author, critical. NICE should beware of placing too much weight on the Birthplace Study.